The major artificial intelligence-focused technology company with the world’s most utilized search engine, Alphabet Inc Class A (Google) (NASDAQ: GOOGL), launched its most advanced AI model on Wednesday.
Gemini 1.0 will be available in three sizes for different users: Ultra, Pro and Nano. Ultra will be capable of highly complex tasks but won’t become widely available until next year, Pro can complete a “wide range of tasks” and Nano is intended for on-device tasks. Google DeepMind was responsible for its development.
Gemini’s Pro and Nano technology is now available in over 170 countries through Google’s chatbot, Bard. The company says it constitutes Bard’s biggest upgrade since it was launched. It is also available through the company’s Pixel 8 Pro smartphone.
“It was built from the ground up to be multimodal, which means it can generalize and seamlessly understand, operate across and combine different types of information including text, code, audio, image and video,” said DeepMind’s Co-Founder and CEO, Demis Hassabis. He says it is the biggest most complicated project Google has ever worked on.
Google says Gemini Ultra is the first AI model to ever outperform human experts on a massive multitask language understanding assessment comprised of 57 different subjects (with a score of 90 per cent). Subjects used to assess the AI bot’s problem-solving capabilities and world knowledge included mathematics, ethics, physics, law, history and medicine.
Gemini will think carefully before answering difficult questions rather than responding based on its first impression of the topic, according to the tech developer.
Read more: Experimental robot glove for stroke patients receives £435k funding from UK government
Read more: Coca-Cola releases limited edition ‘Futuristic’ flavoured soda created with AI
Gemini can quickly filter through and extract insights from thousands of documents
Google says Gemini will be responsible for breakthroughs in multiple fields through its ability to rapidly analyze various types of content and written material.
“Gemini 1.0’s sophisticated multimodal reasoning capabilities can help make sense of complex written and visual information. This makes it uniquely skilled at uncovering knowledge that can be difficult to discern amid vast amounts of data,” says Google.
Gemini has the most comprehensive safety evaluations of any AI model Google has developed. Representatives from the company, alongside other influential AI developers, met with United States President Joe Biden in July this year and pledged to develop AI technologies responsibly.
“We’ve conducted novel research into potential risk areas like cyber-offense, persuasion and autonomy,” said the company.
Google claims it has outperformed OpenAI’s most advanced model, GPT-4, in a series of different tests. These included multiple-choice exams, math tests and other evaluations.
Google has been racing to compete with OpenAI’s platform ever since it was launched about a year ago.
The market value of Google’s parent company Alphabet has risen by approximately 45 per cent this year, representing growth of over $500 billion.
rowan@mugglehead.com
